Automated environmental sound recognition has clear engineering benefits; it allows audio to be sorted, curated, and searched. Unlike music and language, environmental sound is loaded with noise and lacks the rhythm and melody of music or the semantic sequence of language, making it difficult to find common features representative enough of various environmental sound signals. To improve the accuracy of environmental sound recognition, this paper proposes a recognition method based on multi-feature parameters and time–frequency attention module. It begins with a pretreatment that relies on multi-feature parameters to extract the sound, which supplements the phase information lost by the Log-Mel spectrogram in the current mainstream methods, and enhances the expressive ability of input features. A time–frequency attention module with multiple convolutions is designed to extract the attention weight of the input feature spectrogram and reduce the interference coming from the background noise and irrelevant frequency bands in the audio. Comparative experiments were conducted on three general datasets: environmental sound classification datasets (ESC-10, ESC-50) and an UrbanSound8K dataset. Experiments demonstrated that the proposed method performs better.

To reduce the large size of a gunshot recognition network model and to improve the insufficient real-time detection in urban combat, this paper proposes a fast gunshot type recognition method based on knowledge distillation. First, the muzzle blast and the shock wave generated by the gunshot are preprocessed, and the quality of the gunshot recognition dataset is improved using Log-Mel spectrum corresponding to these two signals. Second, a teacher network is constructed using 10 two-dimensional residual modules, and a student network is designed using depth wise separable convolution. Third, the lightweight student network is made to learn the gunshot features under the guidance of the pre-trained large-scale teacher network. Finally, the network’s accuracy, model size, and recognition time are tested using the AudioSet dataset and the NIJ Grant 2016-DN-BX-0183 gunshot dataset. The findings demonstrate that the proposed algorithm achieved 95.6% and 83.5% accuracy on the two datasets, the speed was 0.5 s faster, and the model size was reduced to 2.5 MB. The proposed method is of good practical value in the field of gunshot recognition.

In view of the issues such as the larger network model and lower recognition accuracy of the current gunshot recognition networks, a neural network based on a multi-scale spectrum shift module is proposed in this paper to fully mine the relevant information among the gunshot spectrums. This network employs the architecture of a densely connected convolutional network and uses a multi-scale spectrum shift module on the branch to realize the interaction among spectrum information. This spectrum shift replaces the under-sampling operation among the spectrums, realizes the globalized feature extraction of the spectrum, avoids the loss of information during the under-sampling process, and further improves the quality of the spectrum feature map. Experiments were conducted based on the NIJ Grant 2016-DN-BX-0183 gunshot dataset and YouTube dataset on gunshots that have been open to the public, both of whose classification accuracy reached 83.2% and 95.1%, respectively, with the size of the network model being controlled at around 16 MB. The experimental results indicate that, compared with other existing methods for convolutional neural network, the proposed network can mine globalized time-frequency information better and effectively, and has a higher accuracy of gunshot recognition.

Eddy current (EC) testing demonstrates engineering value for evaluating quality of thermal barrier coatings (TBCs) in aerospace engines and gas turbines, where top ceramic coating (TC) thickness directly determines thermal protection efficiency and component lifespan. Addressing the dual challenges of impedance coupling in multicoating and low sensitivity caused by low conductivity of bond coating (BC) and substrate, this study proposes a method of TC thickness measurement based on high-frequency excitation and near-resonance inductance. A transformer model of EC axial attenuation in TBCs was established, revealing frequency-dependent EC distribution in both BC and substrate. It was demonstrated that increasing the excitation frequency could effectively reduce the interfacial EC density discrepancy, thereby enabling the isolation of TC thickness from coupled impedance. System sensitivity enhancement was achieved through near-resonance impedance amplification, enabled by an equivalent circuit in a targeted frequency band. A mathematical mapping between resonant impedance and TC thickness was established, validating the efficacy of resonant signal enhancement in EC thickness measurement. The experiment of TC thickness measurement demonstrates a tenfold improvement in sensitivity compared to conventional methods, with measurement errors below 4% . This advancement enables precise quality assessment during TBC manufacturing, addressing critical nondestructive evaluation challenges in multicoating systems.

Thermal barrier coatings (TBCs) play a critical role in extending engine service life. Consequently, regular thickness monitoring of TBCs is essential. However, it is difficult to decouple the eddy current (EC) signals of TBCs due to its complex multi-layer structure. In order to accurately measure the coating thickness, a coordinate transformation method based on shape functions is proposed in this paper. Firstly, the simplified mathematical expressions of the characteristics are derived from a transformer model; secondly, the BC and TC signals are successfully decoupled by constructing a characteristic grid and applying a four-point shape function coordinate transformation; subsequently, the optimal combination and frequency are obtained by the evaluation of errors on various characteristic combinations at different frequencies. Finally, an EC detection system is established to evaluate our method. The results demonstrate that the relative errors for BC and TC measurements are less than 6.4% and 4.2%, respectively. Furthermore, this method is effectively immune to the temperature influence on TC thickness detection. Crucially, the approach successfully decoupled the TC and BC layers, with experimental validation confirming high measurement accuracy.

The defects, including cavities, inclusions, and microcracks, are prone to being formed during the directional solidification process of superalloy turbine blades. Eddy current testing (ECT) is challenged by low sensitivity and liftoff noise due to the low conductivity and geometric changes of curved surfaces. An eddy current array (ECA) system is designed in this work, integrating a self-adaptive elastic probe and signal processing algorithms. First, an in-plane differential probe based on the can-shaped magnetic core is developed, where the eddy current density is directionally enhanced through magnetic field shielding to amplify defect response, while liftoff noise is compensated using a self-adaptive elastic structure. Second, an adaptive wavelet threshold denoising algorithm and multichannel fusion algorithm based on array topology prior knowledge enabled effective separation of defect signals from noise. Finally, a hardware system based on multichannel demodulation is developed to improve detection speed and capability. Experimental results demonstrate that the system achieves reliable detection of 0.1 mm defects at a scanning speed of 80 cm/s and achieves a detection probability of 98%. This performance significantly outperforms existing methodologies and provides an effective solution for identifying microdefects in low-conductivity materials, particularly in critical components such as turbine blades.